Almost-Reed-Muller Codes Achieve Constant Rates for Random Errors

This paper considers delta-almost Reed-Muller codes, i.e., linear codes spanned by evaluations of all but a delta fraction of monomials of degree at most d. It is shown that for any delta > 0 and any epsilon > 0, there exists a family of delta-almost Reed-Muller codes of constant rate that correct 1/2-epsilon fraction of random errors with high probability. For exact Reed-Muller codes, the analogous result is not known and represents a weaker version of the longstanding conjecture that Reed-Muller codes achieve capacity for random errors (Abbe-Shpilka-Wigderson STOC '15). Our proof is based on the recent polarization result for Reed-Muller codes, combined with a combinatorial approach to establishing inequalities between the Reed-Muller code entropies.

Automated summary

The paper investigates a new type of linear code that can correct random errors with high probability under a certain level of data loss. This differs from the performance of traditional Reed-Muller codes.